Wisconsin 2010 Senate True Vote Analysis

Richard Charnin

June 16, 2011

Updated: May 6, 2012

This is a True Vote analysis of the 2010 Wisconsin senate race. The Final 2010 Wisconsin Senate Exit Poll was forced to match the recorded vote (Johnson defeated Feingold by 52-47%). Forcing a match to the recorded vote is standard operating procedure on the part of the exit pollsters. In order to force a match in 2004 and 2008, the exit pollsters had to assume an impossible number of returning Bush voters from the previous election.

The returning voter mix should reflect the previous election True Vote, not the recorded vote, In 2010, the Final WI senate exit poll was again forced to match the recorded vote. In the Final, 49% of the 2010 recorded vote were cast by returning Obama voters and 43% by returning McCain voters. The ratio is consistent with Obama’s 7.5% national recorded vote margin. But a WI True Vote analysis indicates that Obama won by 57-41%.

************************************************************************************************************************

5/6/12 Update:

The unadjusted 2008 state presidential exit polls were not available when this analysis was first posted.

Obama won the unadjusted Wisconsin poll by 63-36% (n=2,545 respondents) - a 63.9% (2-party) share. He had 56.2% recorded.

The exit poll 2.43% margin of error includes a 30% "cluster effect".

Based on the MoE, there is a 97.5% probability that Obama's True Wisconsin vote share exceeded 60.77% = 63.20% - 2.43%.

Using the unadjusted polls as a basis, Feingold won by 53.2-45.6%, a 165,000 vote margin.

But even that is conservative: The calculation uses the final exit poll vote shares that were required to match the recorded vote

The shares were adjusted to favor Johnson (Feingold had just 84% of returning Obama voters, an implausibly low share).

If Feingold had a plausible 89% of Obama voters (all else equal), he would have won by 289,000 votes with a 56% total share.

Furthermore, the model assumes 76% McCain voter turnout in 2010, compared to just 68% of Obama voters.

If just 63% of Obama voters and 81% of McCain voters turned out, then all constant, Feingold wins by 52,000 votes with a 50.6% total share.

	2008			2010		Base Case True Vote
	Exit Poll	Share	Alive	Turnout	Voted	Mix	Feingold	Johnson	Taylor
Obama	1,888,503	63.3%	1,814,851	68.0%	1,234,099	56.8%	84%	15%	1%
McCain	1,065,080	35.7%	1,023,542	76.0%	777,892	35.8%	7%	93%
Other	29,834	1.0%	28,671	72.0%	20,643	1.0%	40%	56%	4%
DNV	-	-	-	-	138,697	6.4%	40%	50%	10%
Total	2,983,417	100.0%	2,867,064	70.9%	2,171,331	Share	53.2%	45.6%	1.3%
						Votes	1,154,832	989,463	27,036
							Margin	165,369
						Recorded
						Share	47.0%	51.9%	1.1%
						Votes	1,020,958	1,125,999	23,473
							Margin	-105,041
		2010	Turnout				Feingold share of
		Obama	McCain				Obama	McCain
		68%	76%				84%	7%
Feingold share of Obama						Obama Turnout
Feingold%	79%	84%	89%		McCain	63%	68%	73%
McCain	Feingold Share				Turnout	Feingold Share
12%	52.1%	55.0%	57.8%		71%	52.1%	54.0%	55.8%
7%	50.3%	53.2%	56.0%		76%	51.3%	53.2%	55.0%
2%	48.6%	51.4%	54.2%		81%	50.6%	52.4%	54.2%
	Feingold Margin					Feingold Margin
12%	119,748	243,158	366,568		71%	119,292	199,146	278,999
7%	41,959	165,369	288,779		76%	85,515	165,369	245,222
2%	-35,830	87,580	210,989		81%	51,738	131,592	211,445

************************ end of update ***********************************

It is logical to assume that there were more returning Obama voters than indicated in the 2010 WI exit poll. Therefore, the returning Obama voters share of the 2010 electorate was increased. Election fraud may very well have cost Feingold the election. The True Vote analysis indicates that Feingold had an approximate 50.6% vote share and 50,000 vote margin.

Vote shares were not provided for returning third party (Other) voters and new (DNV) voters in the WI Exit Poll. These categories represented 3% and 5% of the total 2010 recorded vote, respectively. In order to match the recorded vote, Johnson must have won these voters by approximately 61-35% and 64-36%, respectively. This is a RED FLAG! Obama won third party voters by 66-20% over McCain. It strains credulity that returning third party voters would switch from Obama to Johnson in those amounts.

Note the SENSITIVITY ANALYSIS which displays vote shares for various returning voter mix assumptions (scenarios) and Feingold’s share of returning McCain voters. Although the Final Exit Poll is always forced to match the recorded vote, the Base Case True Vote scenario assumes the Final Exit Poll vote shares; only the returning voter mix was adjusted. However, it is likely that the vote shares were also adjusted to force the match.

The True Vote Base Case analysis assumes a 60% turnout of living Obama voters and a 70% turnout of McCain voters and a 1.25% annual voter mortality rate. The percentage mix of returning 2008 third-party (other) voters could not have been the 3% indicated in the WI exit poll. That would mean there were 65,000 third-party voters and there were just 44,000. Therefore, the model assigned the 1.5% excess of Other voters to New/DNV (first-time voters and others who did not vote in 2008).

Feingold was a clear winner assuming that the percentage turnout of returning Obama voters was equal or greater than that of McCain voters. But it is important to keep in mind that the WI exit poll gave Feingold just 84% of returning Obama voters. This is extremely implausible. It is difficult to accept the premise that nearly one of six Obama voters defected to Johnson.

A comparison of the exit poll demographic changes from 2004 to 2010 yields some interesting results:

When Decided: Feingold was the winner in this category.

1) Feingold led the 83% of voters who decided a week prior to the election by 51-48%, and won by 50.7-48.5%.

2) He led the 68% of voters who decided a month prior by 51-48%, and won by 50.4-49.0%.

3) He led the 69% of voters who decided a month prior to the election by 51-48%. Vote shares for the last week were not available. Johnson needed 70% of voters who decided in the last week (17% of the total) to match his recorded share.

Gender: In 2004, females comprised 53% total voters; in 2010, they were just 50% of the total.

Age: Voters over 45 comprised 50% of the electorate in 2004 and 62% in 2010.

Party ID: In 2004, Democrats were 35% and GOP 38% of the electorate. In 2010, Democrats were 37%, the GOP had 36%.

Region: Milwaukee County had 16% of voters in both 2004 and 2010. Feingold had 68% in 2004 but just 61% in 2010

Southeast (incl Waukesha) had 19% of voters in 2004 and 27% in 2010. Feingold had 39% in 2004 and 34% in 2010.

Assuming that Feingold had 88% of returning Obama voters (65% turnout) and Johnson had 93% of returning McCain voters (75% turnout), Feingold was still a 50.0-48.8% winner with a 26,000 vote margin.

			97.5%	Turnout	2010
2008 WI	Recorded	Share	Alive	Rate	Voted	Mix	Feingold	Johnson	Other	Feingold	Johnson	Other
Obama	1677	56.2%	1635	65%	1063	49.0%	88%	11%	1%	935	117	11
McCain	1262	42.3%	1231	75%	923	42.5%	7%	93%	0%	65	858	0
Other	44	1.5%	43	70%	30	1.4%	51%	45%	4%	15	13	1
DNV/New				-	154	7.1%	45%	45%	10%	69	69	15
Total	2983	100.0%	2908	69%	2170	100%	50.0%	48.8%	1.3%	1085	1058	27
									Margin:	26

I have included a comparable Oregon analysis. Oregon Senator Wyden, a popular progressive Democrat like Feingold, was re-elected with a 57% share, closely matching Obama’s 56.8% OR share.

Final 2008 National Exit Poll (recorded vote)					2008 National True Vote
Voted 2004	Mix	Obama	McCain	Other	Voted 2004	Mix	Obama	McCain	Other
DNV	13%	71%	27%	2%	DNV	12.4%	71%	27%	2%
Kerry	37%	89%	9%	1%	Kerry	46.8%	89%	9%	1%
Bush	46%	17%	82%	1%	Bush	39.7%	17%	82%	1%
Other	4%	66%	20%	8%	Other	1.1%	66%	20%	8%
Total	100%	52.6%	45.4%	1.4%	Total	100%	57.9%	40.3%	1.2%
2008 Wisconsin Presidential Exit Poll (recorded vote)					2008 Wisconsin Presidential True Vote
Voted 2004	Mix	Obama	McCain	Other	Voted 2004	Mix	Obama	McCain	Other
New/DNV	7.0%	77.2%	20.9%	1.9%	New/DNV	7.0%	75.5%	22.6%	1.9%
Kerry	46.3%	91.0%	7.1%	1.9%	Kerry	49.3%	89.0%	9.1%	1.9%
Bush	45.8%	17.4%	81.7%	1.0%	Bush	42.8%	17.0%	82.1%	0.9%
Other	0.9%	73.7%	24.4%	1.9%	Other	0.9%	72.0%	26.1%	1.9%
Total	100%	56.2%	42.3%	1.5%	Total	100%	57.1%	41.4%	1.5%
Final 2008 Oregon Exit poll (recorded vote)					2008 Oregon True Vote
Voted 2004	Mix	Obama	McCain	Other	Voted 2004	Mix	Obama	McCain	Other
New/DNV	3%	77%	19%	4%	New/DNV	3.9%	76.2%	20.2%	3.6%
Kerry	51%	90%	6%	4%	Kerry	52.2%	89.0%	7.4%	3.6%
Bush	45%	17%	81%	2%	Bush	42.5%	17.0%	81.2%	1.8%
Other	1%	73%	24%	4%	Other	1.4%	72.0%	24.4%	3.6%
Total	100%	56.8%	40.4%	2.9%	Total	100%	57.7%	39.5%	2.9%

2010 Oregon Senate

Changed exit poll mix from Other (6%) and DNV (5%)

Estimated vote shares to force a match to the recorded vote.

Oregon	Recorded	Votes	Alive	Turnout	Voted	Mix	Wyden	Huffman	Other
Obama	56.75%	1,074	1047	72.8%	763	53%	88%	9%	3%
McCain	40.40%	719	701	73.9%	518	36%	19%	80%	1%
Other	2.85%	53	52	55.7%	29	2%	35%	40%	5%
DNV					130	9%	35%	55%	10%
Total	100%	1846	1800	72.7%	1439	100%	57.3%	39.3%	3.4%
						Votes	825	566	48

2010 Wisconsin Senate

True Vote Base Case

			97.5%	Turnout	2010
2008 WI	True	Share	Alive	Rate	Voted	Mix	Feingold	Johnson	Other	Feingold	Johnson	Other
Obama	1700	57.0%	1658	60%	995	45.8%	88%	11%	1%	875	109	10
McCain	1238	41.5%	1207	70%	845	38.9%	7%	93%	0%	59	786	0
Other	45	1.5%	44	70%	31	1.4%	50%	50%	0%	15	15	0
DNV/New				-	300	13.8%	50%	50%	0%	150	150	0
Total	2983	100.0%	2908	64%	2170	100%	50.7%	48.9%	0.5%	1100	1060	10
									Margin:	39
						Recorded	47.0%	51.9%	1.1%	1020	1126	24

		Feingold share						Feingold share
Obama						Johnson%
Turnout		Feingold % Obama				McCain		Feingold % Obama
50.7%	82%	84%	86%	88%	90%	50.7%	82%	84%	86%	88%	90%
68%	49.9%	50.9%	52.0%	53.0%	54.0%	89%	49.5%	50.4%	51.3%	52.2%	53.2%
66%	49.4%	50.4%	51.4%	52.4%	53.4%	90%	49.1%	50.0%	50.9%	51.8%	52.8%
64%	48.9%	49.9%	50.9%	51.8%	52.8%	91%	48.7%	49.6%	50.5%	51.5%	52.4%
62%	48.4%	49.4%	50.3%	51.3%	52.2%	92%	48.3%	49.2%	50.1%	51.1%	52.0%
60%	47.9%	48.8%	49.8%	50.7%	51.6%	93%	47.9%	48.8%	49.8%	50.7%	51.6%
		Feingold Margin						Feingold Margin
68%	6	51	96	141	187	89%	(12)	27	67	107	147
66%	(15)	28	72	116	160	90%	(29)	10	50	90	130
64%	(37)	5	48	90	133	91%	(46)	(6)	33	73	113
62%	(59)	(17)	24	65	106	92%	(63)	(23)	16	56	96
60%	(80)	(40)	(0)	39	79	93%	(80)	(40)	(0)	39	79

Exit Poll Comparison

	2004				2010
Gender	Mix	FeIngold	Michels		Mix	Feingold	Johnson	Other	Mix Chg
Male	47%	52%	48%		50%	42%	56%	2%	3%
Female	53%	58%	42%		50%	52%	48%	0%	-3%
Total	100%	55.2%	44.8%		100%	44.7%	48.5%	0.8%
Race
White	90%	53%	47%		90%	44%	55%	1%	0%
Black	5%	83%	17%		4%	86%	14%	0%	-1%
Latino	2%	*	*	N/A est	4%	67%	33%	N/A	2%
Asian	1%	*	*	N/A est	1%	67%	33%	N/A	0%
Other	2%	*	*	N/A est	1%	67%	33%	N/A	-1%
Total	100%	51.9%	43.2%		100%	47.1%	52.0%	0.9%
Gender / Race
White Men	43%	50%	49%		45%	40%	59%	1%	2%
White Women	47%	56%	44%		45%	49%	50%	1%	-2%
Non-White Men	4%	69%	30%		4%	70%	30%	0%	0%
NW Women	6%	78%	20%		6%	70%	30%	0%	0%
Total	100%	55.3%	44.2%		100%	47.1%	52.1%	0.9%
Age
18-29	20%	56%	42%	18-29	15%	53%	46%	1%	-5%
30-44	30%	51%	49%	30-44	23%	43%	56%	1%	-7%
45-59	29%	57%	42%	45-64	45%	47%	51%	2%
60+	21%	56%	44%	65+	17%	46%	54%	0%
	100%	54.8%	44.5%	Total	100%	46.8%	51.9%	1.3%
Income
Under $30,000	26%	60%	39%		20%	58%	40%	2%	-6%
$30-50,000	25%	57%	43%		22%	48%	51%	1%	-3%
$50-75,000	24%	55%	45%		27%	46%	54%	N/A	3%
$75-100,000	12%	48%	51%		16%	45%	55%	N/A	4%
$100-200,000	10%	50%	50%		15%	37%	62%	1%	5%
$200,000 +	3%	38%	62%		N/A	N/A	N/A	N/A	na
Total	100%	55.0%	44.7%		100%	47.3%	51.9%	0.8%
Education
No H.S.	5%	65%	35%		3%	44%	55%	1%	-2%
H.S. Graduate	25%	52%	48%		24%	44%	55%	1%	-1%
Some College	34%	55%	44%		35%	43%	56%	1%	1%
College Graduate	22%	51%	48%		24%	47%	53%	0%	2%
Postgraduate	14%	62%	38%		14%	63%	37%	0%	0%
Total	100%	54.9%	44.6%		100%	47.0%	52.4%	0.6%
College Graduate
Yes	36%	56%	44%		38%	53%	47%	0%	2%
No	64%	54%	45%		62%	44%	55%	1%	-2%
Total	100%	54.7%	44.6%		100%	47.4%	52.0%	0.6%
Party ID
Democrat	35%	93%	7%		37%	92%	8%	0%	2%
Republican	38%	14%	86%		36%	5%	94%	1%	-2%
Independent	27%	62%	37%		27%	43%	56%	1%	0%
Total	100%	54.6%	45.1%		100%	47.5%	51.9%	0.6%
Ideology
Liberal	20%	86%	13%		21%	89%	10%	1%	1%
Moderate	48%	63%	37%		43%	58%	42%	0%	-5%
Conservative	32%	22%	78%		36%	11%	89%	0%	4%
Total	100%	54.5%	45.3%		100%	47.6%	52.2%	0.2%
Labor Union
Yes	28%	66%	33%		26%	59%	40%	1%	-2%
No	72%	50%	49%		74%	44%	55%	1%	2%
Total	100%	54.5%	44.5%		100%	47.9%	51.1%	1.0%
Voted in 2008
Obama	na				49%	84%	15%	1%
McCain	na				43%	7%	93%	0%
Other	na			N/A est	3%	36%	54%	10%	Forced
Did not vote	na			N/A est	5%	34%	61%	5%	to match
Total	na				100%	47.0%	52.0%	1.0%
When Decided
Last 3 Days	10%	68%	32%	N/A est	10%	26%	71%	3%	Forced
Last Week	2%	68%	32%	N/A est	7%	33%	66%	1%	to match
Last Month	11%	61%	39%		14%	49%	51%	0%
Before	77%	52%	48%		69%	51%	48%	1%
Total	100%	54.9%	45.1%		100%	47.0%	52.0%	1.1%
When Decided
Last Week	na				17%	49%	51%	N/A
Before	na				83%	51%	48%	1%
Total	na				100%	50.7%	48.5%	0.8%
When Decided
Last Month	na				32%	49%	51%	N/A
Before	na				68%	51%	48%	1%
Total	na				100%	50.4%	49.0%	0.7%
Size of Place
Big Cities	6%	81%	18%		9%	72%	26%	2%	3%
Smaller Cities	19%	61%	39%		6%	69%	31%	N/A	-13%
Suburbs	40%	51%	48%		49%	43%	56%	1%	9%
Small Towns	9%	51%	48%	N/A est	4%	47%	53%	N/A	-5%
Rural	26%	51%	49%		32%	42%	57%	1%	6%
Total	100%	54.7%	44.8%		100%	47.0%	52.0%	1.0%	0%
Size of Place
Urban	26%	66%	34%		16%	71%	28%	1%	-10%
Suburban	39%	51%	48%		49%	43%	56%	1%	10%
Rural	35%	51%	48%		35%	42%	57%	1%	0%
Total	100%	54.9%	44.4%		100%	47.1%	51.9%	1.0%	100%
Region
Milwaukee Cty	16%	68%	32%		16%	61%	38%	1%	0%
Southeast	19%	39%	61%		27%	34%	66%	0%	8%
Southwest	25%	61%	39%		18%	61%	38%	1%	-7%
Northeast	17%	49%	50%		17%	42%	57%	1%	0%
Northwest	23%	57%	42%		22%	46%	53%	1%	-1%
Total	100%	55.0%	44.6%		100%	47.2%	52.1%	0.7%